Highly selective catalytic reduction of NO by H2 over Au0 and Au(I) impregnated in NaY zeolite catalysts

Article abstract of DOI:10.1006/jcat.1996.0274

The NO-H₂ reaction has been studied over Au⁰/NaY and Au(I)/NaY catalysts, which were prepared by the mechanical mixing of AuCl₃ with NaY zeolite and by the monolayer dispersion of Au(I) species into NaY zeolite, respectively. The precursor samples were reduced in flowing hydrogen at 423 K. The NO-H₂ reaction over these catalysts took place at relatively low temperatures under isothermal conditions from 373 to 673 K. The Au(I)/NaY catalysts were more active toward the above reaction compared with the Au⁰/NaY catalysts. The NO selectivity was pointed to the N₂ and N₂O formation, with the former catalysts being more selective to N₂. In situ FT-IR study of the NO-H₂ reaction on the gold catalysts demonstrated that NH₃ was detected in the adsorbed NH₄⁺ form at 1440 cm^-1, along with N₂O at 2240 (2220) cm^-1 and H₂O at 1645 cm^-1. Two characteristic bands due to the vibrations of N-O bond in adsorbed NO were identified. The bands at 1880 cm^-1 in the spectra of Au⁰/NaY were assigned to Auⁿ⁺-NO (n ≈ 0) complexes, and the corresponding bands at 1910 cm^-1 in the spectra of Au(I)/NaY were assigned to Auⁿ⁺ NO (n = 1) complexes in cages of zeolite. It was proposed that N₂O and NH₃ formed simultaneously at low to moderate temperatures (<500 K) by the addition of N atom to NO and H atoms, respectively. The presence of adsorbed N atoms was confirmed by the coadsorption of NO/CO/H₂ mixture onto Au(I)/NaY at 423 K, where a band at 2280 cm^-1 ascribed to isocyanate [NCO] intermediate was developed intensively with time. This band did not deliver when a mixture of NOTCO was admitted alone to the IR cell. Therefore, the H atom promoted the N-O bond fission to form N_ad and OH_ad. The [NCO] species is the result of a reaction between the adsorbed N atom and gaseous CO.

Full text of DOI:10.1006/jcat.1996.0274

JOURNAL OF CATALYSIS 162, 169–178 (1996)
ARTICLE NO. 0274
Highly Selective Catalytic Reduction of NO by H2 over Au0 and Au(I)
Impregnated in NaY Zeolite Catalysts
Tarek M. Salama,1 Ryuichiro Ohnishi, Takafumi Shido, and Masaru Ichikawa2
Catalysis Research Center, Hokkaido University, Sapporo 060, Japan
Received March 20, 1995; revised April 4, 1996; accepted April 8, 1996
out by Galvagno and Parravano (2), supported gold on
SiO2, Al2O3, and MgO catalysts were found to catalyze the
reduction of NO with H2. Lee and Schwank (3) followed up
The NO–H2 reaction has been studied over Au0/NaY and
Au(I)/NaY catalysts, which were prepared by the mechanical mix-
ing of AuCl3 with NaY zeolite and by the monolayer dispersion of on this previous study and used in situ IR for probing cata-
Au(I) species into NaY zeolite, respectively. The precursor samples
were reduced in flowing hydrogen at 423 K. The NO–H2 reaction
over these catalysts took place at relatively low temperatures under
isothermal conditions from 373 to 673 K. The Au(I)/NaY catalysts
were more active toward the above reaction compared with the
Au0/NaY catalysts. The NO selectivity was pointed to the N2 and
N2O formation, with the former catalysts being more selective to
N2. In situ FT–IR study of the NO–H2 reaction on the gold cata-
lysts demonstrated that NH3 was detected in the adsorbed NH4+
form at 1440 cmꢀ1, along with N2O at 2240 (2220) cmꢀ1 and H2O
at 1645 cmꢀ1. Two characteristic bands due to the vibrations of
N–O bond in adsorbed NO were identified. The bands at 1880 cmꢀ1
in the spectra of Au0/NaY were assigned to Aun+–NO (n ꢁ 0) com-
plexes, and the corresponding bands at 1910 cmꢀ1 in the spectra
of Au(I)/NaY were assigned to Aun+ NO (n = 1) complexes in cages
of zeolite. It was proposed that N2O and NH3 formed simultane-
ously at low to moderate temperatures (<500 K) by the addition
of N atom to NO and H atoms, respectively. The presence of ad-
sorbed N atoms was confirmed by the coadsorption of NO/CO/H2
mixture onto Au(I)/NaY at 423 K, where a band at 2280 cmꢀ1 as-
cribed to isocyanate [NCO] intermediate was developed intensively
with time. This band did not deliver when a mixture of NO/CO was
admitted alone to the IR cell. Therefore, the H atom promoted the
lyst surfaces. In agreement with the findings of Galvagno
and Parravano, both Au/SiO2 and Au/MgO were found
active for the reaction between NO and excess H2 at tem-
peratures higher than 573 K. Nevertheless, their IR spectra
quality was rather poor to be informative as the spectra
were collected at high temperatures. A comment must also
be made concerning the high reaction temperature applied
and stabilizing high dispersion of gold phase on the oxide
surfaces since the sublimation energy of gold is relatively
low (4).
It was suggested that further progress could be made by
extending the investigations of the NO–H2 reaction on Au
catalysts by changing the nature of the support. Recently,
we have reported that Au(I)/NaY catalystswere found to be
active in the chemisorption of CO and NO and even surpris-
ingly active in the direct decomposition of NO to N2 and O2
(5, 6). The NO decomposition reaction proceeded through
the sequential formation of N2O3 and N2O intermediates
on Au(I) sites at 300–673 K (5). However, the steady-state
activities were not obtained. It is generally presumed that
the strongly bound oxygen to the catalyst surface prevents
N–O bond fission to form Nad and OHad. The [NCO] species is the further decomposition of NO (7, 8). Thus, it became evi-
result of a reaction between the adsorbed N atom and gaseous CO.
dent that steady-state NO decomposition takes place in the
presence of some gas-phase reductants to remove an excess
oxygen on the catalyst surfaces (9).
c
ꢂ 1996 Academic Press, Inc.
We report here the results of NO reduction by H2 on
Au/NaY as an effort to emphasize the catalyst preparation
on both activity and selectivity of gold catalysts for this re-
action. Although several investigations of NO reduction by
H2 have been reported in the literature, a comprehensive
IR study of this reaction was not provided. Therefore, and
in view of the current catalysts, an in situ infrared spec-
troscopic investigation of NO + H2 reaction is required. A
temperature-programed desorption (TPD) experiment was
carried out to study the affinity of NO decomposition on the
gold catalysts.
1. INTRODUCTION
Au is well known for its inertness among group Ib metals
(1), so that on conventional support surfaces, the Au atoms
may be considerably less active than precious atoms for ad-
sorption and catalysis. However, in a kinetic study carried
1
Permanent address: Chemistry Department, Faculty of Science,
Al-Azhar University, Cairo, Egypt.
2
To whom correspondence should be addressed.
169
0021-9517/96 $18.00
c
Copyright ꢀ 1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.

170
SALAMA ET AL.
YN
2. EXPERIMENTAL
2.1. Catalyst Preparation
2
SN % =
ꢂ 100
2
(YN + YN O
)
)
2
2
Y
N2O
SN O % =
ꢂ 100.
2
Two series of Au (13 wt% ) catalysts were prepared by ap-
plication into NaY zeolite and were noted as Au0/NaY and
Au(I)/NaY to refer to their oxidation states. Since Au(I)
sites are crucial for the NO + H2 reaction, we used this cata-
lyst code to emphasize the moiety of the active gold centers.
However, the Au oxidation state 1 is not the prevailing state
in Au(I)/NaY but it is a mixture of Au(I) and Au0 states,
while in Au0/NaY a metallic Au is the only existing state.
The Au0/NaY samples were prepared at room temperature
by the mechanical mixing of AuCl3 (Strem Chemicals, 99%
purity) with partially dehydrated NaY zeolite (Si/Al = 5.6,
surface area = 910m2 gꢁ1) at 473K. The Au(I)/NaY samples
were prepared by allowing the physically mixed AuCl3/NaY
sample to be exposed to vacuum at 338K for 72h. The AuCl3
vapor migrated to be hydrolyzed by trace H2O in the zeo-
lite. Characterization by XRD, XPS, EXAFS/XANES, and
UV-Vis had proved that Au(III) was homogeneously dis-
tributed in NaY zeolite and was partially reduced to Au(I)
(10, 11).
(YN + YN O
2
2
2.3. In Situ FT-IR (Fourier Transform Infrared)
Spectroscopic Study
The powdered catalyst (15–20 mg) was pressed into a pel-
let and placed in an IR cell equipped with NaCl windows.
The IR cell (dead volume = 94.8 cm3) was connected to a
circulation manifold (dead volume = 402.2 cm3) equipped
with gas supply reservoirs. NO gas was purified prior to
use by passing through a molecular sieve trap cooled in a
slurry of dry ice, acetone, and small amount of liquid N2.
This procedure served to remove NO2 and N2O. Hydrogen
(99.99% ) was used without further purification. A typical
experiment would begin with evacuation of the IR cell at
423 K to remove the moisture and the volatile impurities
from the catalyst pellet. The pellet was pretreated at 423 K
in 50 Torr of H2 for 2 h. A liquid-N2 trap was installed dur-
ing the static reduction of the catalyst pellet to prevent the
evolved H2O from reoxidizing the pellet. A mixture of NO
and H2 (H2/NO = 4), with a total pressure of 68 Torr, was
introduced to the catalyst pellet, so that in situ infrared spec-
tra can be taken of species adsorbed on the catalyst surface.
All IR spectra were collected on a Shimadzu FTIR-4100
spectrometer at a spectral resolution of 2 cmꢁ1 with the
coaddition of 50–100 interferograms.
2.2. Activity Measurements
The activity measurements studies were made at atmo-
spheric pressure using a mixture of 4810 ppm of NO and
1% of H2 balanced with He, in a steady-state flow mode. A
Pyrex glass tubular reactor containing 200 mg of the sam-
ple was used in these runs. Catalytic experiments were done
at equal contact times W/F = 0.47 g s cmꢁ3, where W is the
catalyst weight and F flowrate in millilitersper second. Prior
to each series of runs, the samples were reduced by flow-
ing hydrogen (10 ml minꢁ1) at 423 K for 1 h. The NO and
products in the gas stream at the reactor outlet were contin-
uously analyzed using on-line gas chromatograph (Ohkura
GC-103) with thermal conductivity detector. Two separa-
tion columns were used. The first column was packed with
porapak-N (4 m long, ꢀ = 4 mm) for N2O and NH3 separa-
tion, while the other was packed with 5A molecular sieve
(2 m long, ꢀ = 4 mm) for N2 and NO separation. He was
used as a carrier gas with flow rates equal to 25 ml minꢁ1
for the former column and 15 ml minꢁ1 for the latter. The
columns temperature was 363 K. The NO conversion has
been evaluated as the percentage of NO converted to re-
duced products i. e., N2 and N2O. The selectivity, S, for N2
and N2O is calculated from the corresponding yield, Y, as
follows:
2.4. TPD Study
TPD experiments were conducted in a fixed-bed reactor
using 50 mg of the sample. NO gas (5 Torr) was admitted
onto the catalyst bed at 300 K for 15 min, after which time
the system had equilibrated. Following this, the system
was evacuated until a base pressure of 2 ꢂ l0ꢁ7 Torr was
attained. TPD started by increasing the temperature
linearly with the rate of 5 K minꢁ1 and scan time 10 s. The
desorbed gases were analyzed by ANELVA AQA-100
quadrupole mass spectrometer and gave the following
elution traces: NO (mass 30), N2O (mass 44), N2 (mass 28),
and O2 (mass 32).
3. RESULTS
3.1. Activity versus Time and Temperature
Figure 1 shows the NO conversion to reduced products in
the NO–H2 reaction, i.e., N2 and N2O, as a function of time-
on-stream after various reaction temperatures were applied
to Au(I)/NaY catalysts. Interestingly, the reaction began at
relatively lower temperatures, such as 373 K. All the conver-
sions were in the steady state and did not decay for hours.
The specific ratesfor NO reduction to N2 and N2O under the
steady-state conditions over the gold catalysts as a function
2N2
YN
=
=
2
NOinlet
2N2O
YN O
2
NOinlet

SELECTIVE REDUCTION OF NO BY H2 ON Au0/NaY AND Au(I)/NaY
171
FIG. 1. Conversion of NO to N2 + N2O as a function of time-on-
stream in the NO–H2 reaction over Au(I)/NaY at various reaction tem-
peratures.
FIG. 2. Comparison of the NO selectivity to N2 and N2O on the gold
catalysts as a function of reaction temperatures.
catalytic activity variations in Table 1, we found that the
subsequent measurements of rates at lower temperatures
following the reaction at 673 K reproduced the original val-
ues. Therefore, such variationsare related to thermodynam-
ically favorable products distribution as shown below.
of reaction temperature are compared in Table 1. The rates
of N2 and N2O formation on Au(I)/NaY are almost twice
as high as those on Au0/NaY in the temperature range
433–623 K, and the difference in rates increased almost five
times at 673 K. There is a common rate trend accompany-
ing the two catalysts in the temperature range 373–623 K.
The specific rates first increased to 433 K, where substan-
tial amounts of N2O were detected, followed by a drop at
433–473 K and passed through a very broad minimum val-
ues for temperatures between 473 and 573 K. The specific
catalytic activities of both catalysts were increased by in-
creasing the reaction temperature above 573 K, i.e., 623 K.
As the reaction temperature is further increased to 673 K,
the rate of NO reduction increased again with Au(I)/NaY,
while it sharply decreased with Au0/NaY. On testing the
structure stability of the catalysts that might result in the
3.2. Products Distribution versus Reaction Temperature
Figure 2 shows the selectivity of NO in the NO–H2
reaction on gold catalysts versus the temperature. It should
be stressed that distribution of products N2 and N2O de-
pends on the reaction temperature. The NO selectivity to
N2O and N2 is 100% in the range of temperatures applied
in this study, i.e., 373–673 K. No NH3 is ever detected at the
outlet, although this clearly counts against the results of
Galvagno and Parravano (2) for the same reaction on Au
supported on Al2O3, SiO2, and MgO catalysts. The prod-
uct distributions are somewhat similar in behavior on both
types of catalysts. N2O was produced as an important prod-
uct at 433 K. By increasing the reaction temperature, the
NO selectivity to N2O drops steeply on both Au0/NaY and
Au(I)/NaY catalysts with corresponding increase in the se-
lectivity to N2. The selectivity to N2 was 97% on Au(I)/NaY
at 673 K and 82.4% on Au0/NaY at 673 K.
TABLE 1
Rates of Nitric Oxide Reduction by Hydrogen on the Gold
Catalysts at Various Temperatures
Rate (10ꢁ2 ꢃ mola NO/gb-Au ꢃ min)
Temp. (K)
Au(I)/NaY
Au/NaY
4. IN SITU FT-IR STUDY OF THE NO–H2 REACTION
373
423
474
573
623
673
1.08
4.35
3.04
3.22
4.49
6.69
1.01
2.93
2.00
1.95
3.89
1.45
The effects of the reaction temperature against time on
the infrared species observed during the NO–H2 reaction
over Au0/NaY and Au(I)/NaY are shown in Figs. 3–7 . The
spectra obtained were subtracted from background for each
spectrum. All the bands seen in these spectra are due to vi-
brations of adsorbed N2O at 2240 (2220) cmꢁ1 (12), and NO
in Au–nitrosyl complexes: Aun+–NO (n = 1) at 1910 cmꢁ1
(13, 14), Aun+–NO (n ꢄ 0) at 1880 cmꢁ1 (13, 14), Au–NOꢁ
aMoles of NO that have been converted to N2O and N2.
bAu content was determined by ICP analysis. Rates were normalized
to a total gram gold content in the samples.

172
SALAMA ET AL.
the H2O bands at 1645 cmꢁ1. It is noteworthy to mention
that the bands due to monomeric Au–nitrosyl complexes at
1880 and 1800 cmꢁ1 are seen only in the spectra of Au0/NaY
at 433K. Therefore, NO did not completelydissociate at this
temperature. These bands conceivably could be assigned to
adsorbed NO in Aun+–NO (n ꢄ 0) complexes by analogy
with the NO band on Ag+/NaY and on Cu+/ZSM-5 (13).
From the observed gradual peak shift to lower frequency
with reaction time, i.e., from 1890 to 1880 cmꢁ1, it is con-
cluded that the NO bond order is weakened due to charge
transfer from Aun+ (n ꢄ 0) to NOꢁ+. The charge donation
from Aun + (n ꢄ 0) to NOꢁ+ centers is almost complete as
the observed IR frequency is slightly higher than that of
free NO at 1876 cmꢁ1. The 1800 cmꢁ1 band is likely to be
assigned to NOꢁ vibration.
When the NO–H2 reaction temperature was raised to
473 K, the spectra obtained resembled those collected over
a Au0/NaY catalyst at 433 K except that the bands at 1880
and 1800 cmꢁ1 have completely vanished, as shown in Fig. 4.
Thus, the slightly upscale shift of the band at 1880 cmꢁ1
from that of gaseous NO indicates that the structure con-
tributing this bond is weakly attached on the surface and
is rapidly removed by increasing the temperature. Conse-
quently, it appears more reasonable to propose that this
form of adsorbed NO and the NO species at 1800 cmꢁ1 are
FIG. 3. In situ FT-IR spectra from NO–H2 reaction on Au0/NaY at
433 K. The sample was exposed to reactant gases for (a) 5 min, (b) 15 min,
(c) 30 min, (d) 45 min, (e) 75 min, (f) 90 min, and (g) 120 min.
at 1800 cmꢁ1 (15). H2O has a band at 1645 cmꢁ1 (16) and
adsorbed NH+4 at 1440 cmꢁ1 (16). No form of adsorbed
NH3 molecules at 1620 cmꢁ1 was detected. The hydrogen
bonded to surface hydroxyl, caused by H2O, masked the
NH bands in the high-frequency regime beyond recogni-
tion. The detection of adsorbed H2O species upon the ad-
mission of NO–H2 mixture onto the gold samples at low
to moderate temperatures (<423 K) is a crucial evidence
that N2O is not the result of NO dissociation but is a re-
duced product. Once the bands of product species N2O and
NH+4 formed, their intensities gradually increased with time
to reach certain maxima then showed no further changes.
Thisindicatesthat the reaction equilibrium wasattained at a
given time, and furthermore, these species did not undergo
further reaction.
4.1. Au0/NaY
As shown in Fig. 3, when the Au0/NaY catalyst was con-
tacted with a mixture of NO–H2 at 433 K, the bands of
FIG. 4. In situ FT-IR spectra from NO–H2 (H2/NO = 4) reaction
product species N2O and NH+ were produced, along with (a) 5 min, (b) 15 min, (c) 30 min, (d) 45 min, (e) 60 min, and (f) 120 min.
on Au0/NaY at 473 K. The sample was exposed to reactant gases for
4

SELECTIVE REDUCTION OF NO BY H2 ON Au0/NaY AND Au(I)/NaY
173
the ones that dissociate. The relative peaks intensity ratio
NH+4 /H2O decreased from 8 to 5.3 by raising the tempera-
ture and became invariant with time after 30 min. Thus, the
concentration of NH+4 species decreased with raising the
temperature from 433 to 473 K.
4.2. Au(I)/NaY
Shown in Fig. 5 are the IR spectra of NO–H2 reaction
on Au(I)/NaY at 373 K. It is interesting to note that the
bands of reduced species N2O and NH+4 were produced at
this relatively low temperature. The Au–NO bands were
broad and ill defined. The N–O bond order at 1910 cmꢁ1
became stronger by the positive frequency displacement
above the vibrational frequency of free NO than the corre-
sponding one at 1890–1880 cmꢁ1 in the spectra of Au0/NaY
(Fig. 3). The 1910 cmꢁ1 band would thus be assigned to a
partially positive NO species on Au(I)/NaY. However, it is
conceivable that the removal of one antibonding electron
by coordinating to Au(I) site is contributing bonding mode
to NO at 1910 cmꢁ1. The weak bands at about 1850 cmꢁ1 are
possibly assigned to NO vibrations in Au–NOꢁꢁ structure.
Figure 6 shows the IR spectra of NO–H2 reaction over
Au(I)/NaY at 433 K. By comparing these spectra with those
in Fig. 5, the changes in the Au(I)-nitrosyl features were not
significant. Since the band at 1910 cmꢁ1 did not diminish at
this temperature as the 1890–1880 cmꢁ1 band did in the
spectra of Au0/NaY at the same temperature, it must be
noted that the NO molecule is accommodated as a pos-
FIG. 6. In situ FT-IR spectra from NO–H2 (H2/NO = 4) reaction
on Au(I)/NaY at 433 K. The sample was exposed to reactant gases for
(a) 5 min, (b) 10 min, (c) 20 min, (d) 45 min, (e) 60 min, (f) 90 min, (g) 120
min, and (h) 180 min.
itively charged state in the sodalite cavity in Au(I)/NaY.
Weak IR bands characteristic of Au(NO)2 species at 1817
and 1736 cmꢁ1 could be detected. The bands for dinitrosyl
species are analogous to those on Cu/ZSM-5 (13). How-
ever, it is a relatively less important species under our re-
action conditions. On the other hand, the data shown in
Fig. 6 demonstrate pronounced differences in the intensity
ratios NH+4 /H2O compared to the analogies in Fig. 3. These
ratios varied from 1.82 to 0.94 over 2 h reaction time on
Au(I)/NaY, while it varied from 0.8 to 0.66 over the same
time on Au0/NaY. Therefore, the activation ofmolecular hy-
drogen is facilitated better on Au(I)/NaY than on Au0/NaY
under the same experimental conditions.
Upon increasing the NO–H2 reaction temperature to
573 K, the bands responsible for both N2O and NH+4 species
were not produced (Fig. 7). Also, the Au(I)–NO bands have
not developed. The H2O band intensity decreased due to
the intense scattering of the infrared beam by the sample
and/or desorption of H2O from the sample at high tem-
peratures. To check up the generation of N2O and NH4+
FIG. 5. In situ FT-IR spectra from NO–H2 (H2/NO = 4) reaction on
Au(I)/NaY at 373 K. The sample was exposed to reactant gases for
(a) 5 min, ( b) 10 min, (c) 20 min, (d) 30 min, (e) 45 min, (f) 60 min, (g) 80
min, (h) 105 min, (i) 120 min, (j) 150 min, (k) 175 min, (l) 210 min, and
(m) 240 min.

174
SALAMA ET AL.
the CO adsorption on Au/MgO gave a band that was shifted
to a significantly lower wavenumber in comparison to the
band position on Au/SiO2. Lee and Schwank (3) plausibly
concluded that an interaction of gold with oxide supports
can, under certain conditions, modify the chemistry of sup-
ported gold, and the active sites in Au/SiO2 and Au/MgO
for NO–H2 reaction are not necessarily zero-valent metallic
gold. This suggests that the presence of a small fraction of
Aun+ in the gold catalysts is necessary in the activation of
molecular H2 as well as NO.
The data in Table 1 and Fig. 2 showed that the Au(I)/NaY
catalysts are more active and selective than the Au0/NaY
catalysts in the NO–H2 reaction, especially at 673 K. This
is due to the presence of charged Au cations in pores of
the former catalysts since larger Au particles are expected
to be segregated and exposed to external zeolite at the
prevailing temperatures, thereby leading to the activity de-
crease. Thus, the dispersion state of Au atoms is related to
changes of the chemical states of these atoms. The presence
of charged Au atoms was confirmed by the in situ IR spec-
tra of the NO–H2 reaction. From the relative NO band shift
above that of free NO, it is concluded that stronger charged
Au atoms were generated in Au(I)/NaY preparations than
in Au0/NaY preparations. The charged Au atoms are ex-
pected to be located in the pores of NaY zeolite in the for-
mer catalysts for the following reasons: First, their charac-
teristicNO bandsat 1910cmꢁ1 did not diminish at 473K. On
the contrary, the corresponding bands at 1890–1880 cmꢁ1 in
the spectra of Au0/NaY disappeared completely at 423 K.
Second, according to the IR data the reaction equilibrium
was attained on Au0/NaY faster than on Au(I)/NaY (Figs. 3
and 6). Third, the activities of both catalysts at 673 K were
dramatically different.
When a gas mixture of NO/H2/CO was admitted onto an
Au(I)/NaY pellet at 473 K and collected the IR spectra on
cooling the gold catalyst, a characteristic strong carbonyl IR
absorbing band at 2188 cmꢁ1 was detected. This band was
reasonably assigned to a CO vibration of carbonyl coordi-
nated Au(I) site (20). Besides the IR investigation in this
study, we are inclined to consider that Au(I) sites could be
formed in pores of zeolite under the net redox atmosphere
used in the NO/H2 reaction. As the oxidation affinity of NO
is higher than that of H2O as a side product, the latter has a
minimal effect on changing the oxidation state of Au atoms.
Thus, the difference in catalystic activities in Table 1 must be
discussed in terms of the presence of a dual Au(I) and Au0
site in Au(I)/NaY and a unique Au0 site in Au0/NaY, which
is exposed to the surface as big particles especially at high
temperatures (Table 1). However, the Au(I) sites in general
are very remarkable in NO dissociation (5). Au(I) and Au0
must exist next to each other since it is highly unlikely that
Au(I) ions were located neither within the crystal structure
of metallic Au particles nor as isolated small patches. Metal
ions have been proven to be present in macroscopic inter-
FIG. 7. In situ FT-IR spectra from NO–H2 (H2/NO = 4) reaction
on Au(I)/NaY at 573 K. The sample was exposed to reactant gases for
(a) 30 min, (b) 60 min, (c) 105 min, (d) after (c) cooling to 474 K, (e) after
(d) cooling to 423 K, (f) after (e) cooling to 373 K, and (g) after (f) cooling
to 298 K.
species on Au(I)/NaY at 573 K, the spectra were collected
on cooling the sample (Figs. 7e–7g). No bands appeared
at 2240(2220) and 1440 cmꢁ1; the only ones detected were
those for H2O at 1645 cmꢁ1. Therefore, the production of
NH3 and N2O essentially ceased and the selectivity was di-
rected only to N2 at temperatures >573 K, in agreement
with the product distribution data given in Fig. 2. This is in
direct contrast to the catalytic behavior of Pt and Pd cata-
lysts in NO–H2 reaction, where significant NH3 formation
is observed up to temperatures as high as 1033 K (17).
5. DISCUSSION
5.1. Catalytic Performance of Au-Supported Catalysts
in NO–H2 Reaction
Although the dissociation of molecular H2 is rather dif-
ficult on Au (18), it has been reported that the H2–D2 ex-
change reaction proceeded on Au/SiO2 and Au/MgO at the
temperature used in NO–H2 reaction, i.e., 573 K (2). The ac-
tivity of Au/SiO2 was about an order magnitude higher than
that of Au/MgO. In a parallel study by Schwank, et al. (19),

SELECTIVE REDUCTION OF NO BY H2 ON Au0/NaY AND Au(I)/NaY
175
faces between metals and oxidic materials (21, 22). Thus, it
is suggested that Au(I) ions exist at the interface between
metallic Au particles and interior walls of NaY zeolite.
The decline in activity of NO conversion between 433 and
573 K has been previously reported for the same reaction on
Pt/ZSM-5 (23), which was attributed to the lack of available
hydrogen because of the simultaneous formation of NH3.
The NH3 is presumably produced by the addition of hydro-
gen atoms to N atoms derived from NO dissociation. Al-
though, NH3 was not observed in the reaction products over
the gold catalysts due to a H2 dissociation process less effi-
cient than that on Pt, adsorbed NH+4 species was detected
in the IR spectra of the NO + H2 reaction at temperatures
lower than 573 K. It appears likely that small amounts of
NH3 that are converted in the presence of H+ were pro-
duced, from polarized H2O to adsorbed NH+4 species at
1440 cmꢁ1. However, a mass balance of 70 ꢅ 10% on N2
was obtained at temperatures lower than 573 K. This value
has increased to 90 ꢅ 10% by increasing the reaction tem-
perature over ꢄ 573 K. Thus, it was confirmed that the gaps
among the NO conversions to N2O + N2 in the temperature
range 433–573 K are linked by adsorbed NH+4 and N2O
species. Although, the concentration of NH+4 species was
decreased then vanished by raising the temperature to 573
(Figs. 5–7), the presence of strongly adsorbed N2O species
onto Au(I) sites in cages of zeolite has been confirmed from
NO-TPD results in the same temperature region, as is pre-
sented below.
5.2 Reaction Mechanism
FIG. 8. In situ FT-IR spectra from N2O–H2 reaction on Au(I)/NaY
at 433 K as a function of reaction time, (a) 30 min, (b) 60 min, and
(c) 120 min.
5.2.1. At temperatures <573 K. The in situ spectra of
N2O + H2 reaction over Au/NaY at 423 K are shown in
Fig. 8. It is observable that the NH+4 bands cannot be pro-
duced by the above reaction according to the following
equation:
equation:
NHx + NO → N2O + x/2Hx O.
N2O + 4H2 → 2NH3 + H2O.
Recently, it has been reported that there is a direct cor- The subsequent addition of H2 (160 Torr) to the reactor
relation between concentration of NH+4 and NOx conver- at the same temperature caused the development of N2O
sion over vanadia–titania catalysts (24), and NH4+ are active and NH+4 bands with time again (Figs. 9f–9k). This re-
species for the NO–NH3 to produce N2. Figure 9 illustrates sult unequivocally shows that the H2 adsorption is disso-
the spectra resulting from NO + H2 reaction on Au(I)/NaY ciative, and therefore H atoms must have been involved
at 423 K in sequential treatment. After the reaction pro- in a common reaction pathway to produce N2O and NH4
ceeded for 2 h (Fig. 9a), the system was evacuated shortly simultaneously.
for 5 min (Fig. 9b). As a result, the N2O bands have dis-
From the TPD measurements of NO adsorption on
appeared leaving the bands of H2O and NH+4 . Following Au(I)/NaY catalysts (Fig. 10), the associative adsorption of
this, NO (40 Torr) was introduced first for the purpose of NO molecules was deduced in the temperature range where
interacting with adsorbed NH+4 species. As can be seen, the the parallel formation of N2O and NH+4 species was ob-
N2O bands developed to a lesser extent with a negligible served in the IR spectra at temperatures lower than 573 K.
decrease in the concentration of NH+4 species (Figs. 9c–9e). The N2O formation should primarily involve the NO dis-
The conclusion from this result is unlikely to be that the dis- sociation. For the N–O bond to be broken it is anticipated
sociation of adsorbed NH+4 species via series of NHx inter- that it will be necessary to first weaken the bond through
mediates is a major reaction path for producing N2O under the addition of an electron into the antibonding orbital of
these experimental conditions according to the following the N–O bond. According to the spectral features in Fig. 9 it

176
SALAMA ET AL.
(6) NH3,ad + H+(HCl, H2O) → NH+4,ad
(7) Had + OHad → H2O.
Hecker and Bell have postulated (26) that the NO dissoci-
ation proceeds with the assistance of an adsorbed H atom.
We have, however, confidently probed the reaction path-
way 3 by the IR spectra recorded in sequential NO/CO/H2
additions over Au(I)/NaY at 423 K (Fig. 11). The NO/CO
(CO/NO = 4) mixture was introduced first to the catalyst
pellet at 423 K. After 1 h contact time, the interaction
between NO and CO did not produce additional bands
(Fig. 11a). As soon as H2 (CO/H2 = 1) was admitted to the
system, a band at 2280 cmꢁ1 was produced, which systemat-
ically increased in intensity with time (Figs. 11b–11g). This
band is assigned to isocyanate–[NCO] intermediates (27).
It became apparent from this result that a reaction occurs
between NO and H2 to produce surface N atoms, and there-
fore H atoms promoted the NO dissociation according to
step 3, in full agreement with the assumption proposed by
Hecker and Bell (26). Adsorbed N atoms interacted with
gaseous CO (bands at 2176 and 2120 cmꢁ1) to form an iso-
cyanate complex:
Au–N + CO ↔ Au–NCO.
5.2.2. At temperatures >573 K. When the temperature
ofthe Au(I)/NaY catalyst exceeded 573K, the NO molecule
dissociated mainly to produce N2O and N2 regardless of the
H-assistant role, as shown in Fig. 10. The desorption of N2O
FIG. 9. In situ FT-IR spectra of adsorbed species on Au(I)/NaY at
423 K. (a) The sample was exposed to NO (17 Torr) + H2 (51 torr) for
2 h; (b) the sample was evacuated for 5 min after (a); after (b) the sample
was exposed to NO (40 Torr) for (c) 20 min, (d) 60 min, and (e) 180 min;
after (e) the sample was exposed to NO (40 Torr) + H2 (160 Torr) for
(f) 5 min, (g) 15 min, (h) 25 min, (i) 35 min, (j) 50 min, and (k) 1 h.
seems reasonable that an electron from a H atom weakened
the N–O bond, thereby causing to N–O bond dissociation
at these temperatures.
Kokes (25) has suggested that the NO–H2 reaction pro-
ceeds via the formation of (HNO)x, a solid species that de-
composes readily to yield N2O, NH3, and H2O. Our data
agree in general with Kokes’ findings, the following reac-
tion paths seem reasonable, and, in particular, close homo-
geneous analogies to it have developed at low to moderate
temperatures (<573 K):
(1) H2 → 2Had
(2) NOg → NOad
(3) NOad + Had → Nad + OHad
(4) Nad + NOad → N2Og
(5) Nad + 3Had → NH3,ad
FIG. 10. TPD profiles of NO decomposition on Au(I)/NaY to form
N2 (m/e = 28), NO (m/e = 30), and N2O (m/e = 44).

SELECTIVE REDUCTION OF NO BY H2 ON Au0/NaY AND Au(I)/NaY
177
(4) N2Oad + 2Had → N2,g + H2O
(5) 2Had + Oad → H2O.
6. CONCLUSIONS
(1) The NO–H2 reaction took place on gold catalysts at
temperatures as relatively low as 373–673 K. The NO se-
lectivity to dinitrogen-containing products (N2O and N2)
was rather high unless it is 100% especially at tempera-
tures above 500 K. NH3 was not detected at the outlet,
although the band of adsorbed NH+4 at 1440 cmꢁ1 grew in
the in situ FT-IR spectra of the above reaction up to 500 K.
NH3 was produced with small amounts, which react subse-
quently with H2O to produce NH+4 .
(2) The in situ FTIR spectra of NO + H2 reaction on
Au(I)/NaY revealed that the bands of NOx+ coordinating
Aun+ sites appeared at 1910 cmꢁ1, at such a location above
the frequency of free NO at 1876 cmꢁ1. The corresponding
bands from Au0/NaY were located at 1880 cmꢁ1. Therefore,
the band at 1910 cmꢁ1 was assigned to vibrations of NO in
Au(I)–NO complexes in cages of NaY zeolite, while those
at 1880 cmꢁ1 were assigned to NO vibrations in Aun+–NO
(n ꢄ 0) complexes on the external zeolite or near by regions
of the zeolite surface.
(3) The NO-TPD profiles from Au(I)/NaY catalyst re-
vealed the associative adsorption of NO up to 573 K. At
these temperatures, the NO molecule was dissociated to Nad
and OHad bythe aid ofH atom in the NO + H2 reaction. This
interpretation is completely consistent with the observation
of a band at 2280 cmꢁ1 ascribed to adsorbed [NCO] species
when a mixture of NO/CO/H2 was adsorbed on Au(I)/NaY
at 432 K. However, the coadsorption of only NO/CO did
not produce this species. The [NCO] species is formed by a
reaction between adsorbed N atom and gaseous CO.
(4) The dissociative adsorption of NO occurred at
temperatures higher than 573 K. The N2O is proposed as
the intermediate species in the NO + H2 reaction under
the same NO dissociation conditions. The formation of N2
by the decomposition of adsorbed NH+4 species via series
of NHx intermediates was not considered a major reaction
path. This is because the reaction between NO and NH4+
species did not yield the development of N2O bands at
423 K on one hand and the decrease in the NH+4 concen-
tration was not associated with a corresponding increase
in either N2O or N2 compositions on the other hand.
FIG. 11. In situ FT-IR spectra of adsorbed species on Au(I)/NaY at
423 K. (a) The sample was exposed to NO (17 Torr) + CO (51 Torr) for
1 h; then H2 (50 Torr) was introduced to the reaction mixture for (b) 5
min, (c) 20 min, (d) 30 min, (e) 45 min, (f) 90 min, and (g) 120 min.
at 615 K implies the existence of stable Au(I) cations in the
zeoliote cages. The formation of N2O was reported to be
correlated to the existence of Cu(I) on Cu/Al spinel catalyst
in the NO decomposition reaction (28). However, the very
broad N2O desorption peak at this temperature is likely the
result of a diffusion limited process from the sodalite cage
into the gas phase. The temperature at which N2O started to
decay almost coincided with the temperature of increased
NO selectivity to N2, e.g., 573–673 K (Fig. 2). This indicates
that N2O is a reaction intermediate to N2 in the NO–H2 at
temperatures higher than 573 K. The formed N2 desorbs
immediately from the Au sites and new vacancies were cre-
ated. The oxygen atoms left on the surface were removed
from the catalyst by hydrogen to form H2O according to
the subsequent steps:
REFERENCES
1. Anderson, J. R., “Structure of Metallic Catalysts,” pp. 21–25.
Academic Press, New York, 1975
2. Galvagno, S., and Parravano, G., J. Catal. 55, 178 (1978).
3. Lee, J. Y., and Schwank, J., J. Catal. 102, 207 (1986).
4. Puddephatt, R. J., in “The Chemistry of Gold.” (R. J. H. Clark, Ed.),
Elsevier, Amsterdam, 1978.
(1) NOg → NOad
(2) H2,g → 2Had
(3) 2NOad → N2Oad + Oad

178
SALAMA ET AL.
5. Salama, T. M., Shido, T., Ohnishi, Y., and Ichikawa, M., J. Chem. Soc. 16. Little, L. H., in “Infrared Spectra of Adsorbed Species.” Academic
Chem. Commun., 2749 (1994). Press, London, 1966.
6. Qiu, S., Ohnishi, R., and Ichikawa, M., J. Chem. Soc. Chem. Commun., 17. Kobylinski, T. P., and Taylor, B. W., J. Catal. 33, 376 (1974).
1425 (1992).
18. Harrison, B., Wyatt, M., and Gough, K. G., in “Catalysis,” (G. C. Bond
and G. Webb, Eds.), Specialist Periodical Report, Vol. 5, p. 127. Royal.
Chem. Soc., London, 1982.
19. Schwank, J., Parravano, G., and Gruber, H. L., J. Catal. 61, 19 (1980).
20. Salama, T. M., Ohnishi, Y., and Ichikawa, M., J. Chem. Soc., Faraday
Trans. 92, 301 (1996).
7. Iwamoto, M., in “Future Opportunities in Catalytic and Separation
Technology” (M. Misono, Y. Yoro-oka, and S. Kimura, Eds.), p. 121.
Elsevier, Amsterdam, 1990.
8. Li, Y., and Hall, W. K., J. Catal. 129, 202 (1991).
9. Li, Y., and Armor, J. N., Appl. Catal. 76, L1 (1991).
10. Salama, T. M., Shido, T., Minagawa, H., and Ichikawa, M., J. Catal. 21. Salama, T. M., Hattori, H., Kita, H., Ebitani, K., and Tanaka, T., J.
152, 322 (1995). Chem. Soc., Faraday Trans. 89, 2067 (1993).
11. Salama, T. M., Shido, T., Ohnishi, Y., and Ichikawa, M., J. Phys. Chem. 22. Salama, T. M., Ebitani, K., Hattori, H., and Kita, H., Chem. Mater. 6,
100, 3688 (1996).
21 (1994).
12. Chao, C.-H., and Lunsford, J. H. L., J. Am. Chem. Soc. 93, 71
(1971).
13. Valyon, J., and Hall, W. K., J. Phys. Chem. 97, 1204 (1993).
14. Chao, C.-H., and Lunsford, J. H. L., J. Phys. Chem. 78, 1174
(1974).
23. Burch, R., and Scire, T., Catal. Lett. 27, 177 (1994).
24. Topsoe, N.-Y., Science 265, 1217 (1994).
25. Kokes, R. J., J. Phys. Chem. 70, 296 (1966).
26. Hecker, W. C., Bell, A. T., and J. Catal. 92, 247 (1985).
27. Solymosi, F., Sarkany, J., and Schauer, A., J. Catal. 46, 297 (1977).
28. Hierl, R., Urbach, H., and Knozinger, H., J. Chem. Soc., Faraday Trans.
88, 355 (1992).
15. Zhang, W.-X., Yahiro, H., Mizuno, N., Izumi, J., and Iwamoto, M.,
Langmuir 9, 2337 (1993).